DM: exptl. (group) treatments during one primary period?

[Chris Jones]

I've recently run an experiment where I've monitored 6 enclosed prey populations using CMR (4 sessions of 5-9 occassions each). Prey were exposed to a range (5 treatment levels with replication of control = 6 groups) of densities of predators during the final primary period (i.e. between sessions 3 and 4). I plan to use the robust design to investigate a potential linear relationship between predator density and prey survival, but I am unsure of the correct design matrix.

I have two questions:

1. Can I run a robust design model for the entire 4-session data set, yet apply a group covariate to only one time interval, and how do I structure this?

2). I have the option of just using data from the last two sessions (again in a robust design). Is it beter to construct the DM with a separate column for each group (pred. density), or to use one column of continuous covariates to code for this?

Thanks for any helpful suggestions.